把石墨评论导入word

requirements.txt

bayoo-docx==0.2.0

shimo_comment.py

#!/usr/bin/env python
# -*- coding: utf-8 -*-

import json
import docx
import re

with open('chapter2-comments.txt', 'r', encoding='utf-8') as f:
    commtext = json.load(f)
with open('maintext-dedup.txt', 'r', encoding='utf-8') as f:
    maintext = json.load(f)
document = docx.Document("modified.docx")

comment_refs = []
comment_pattern = re.compile(r'\"(comment-\w+)\"')

comments = {}

for cmt in commtext:
    guid = cmt['selection_guid']
    if guid in comments:
        comments[guid].append(cmt['content'].strip())
    else:
        comments[guid] = [cmt['content'].strip()]

for mtxt in maintext:
    if len(mtxt) == 3:
        match = comment_pattern.search(mtxt[2])
        if match is not None:
            comment_refs.append((match[1], mtxt[1].strip()))

i = 0
j = 0
paras = document.paragraphs
while True:
    cmt = comment_refs[j]
    p = paras[i]
    matching = re.search(re.escape(cmt[1]), p.text)
    if matching is None:
        i = i + 1
        if i >= len(paras):
            print(cmt[1])
            print("selection", cmt[1], "is never found!")
            raise ValueError()
    else:
        print("found selection", cmt[1], "at paragraph", i)
        for cmt_text in comments[cmt[0]]:
            p.add_comment(cmt_text)
        j = j + 1
        if j >= len(comment_refs):
            print("finished!")
            break
        
document.save('modified.docx')